Professional Documents
Culture Documents
RESEARCH ARTICLE
Abstract
OPEN ACCESS
Low-effort, reliable diagnostics of digital dermatitis (DD) are needed, especially for lesions
Citation: Vanhoudt A, Jacobs C, Caron M, Barkema
HW, Nielen M, van Werven T, et al. (2023) Broad- warranting treatment, regardless of milking system or hygienic condition of the feet. The pri-
spectrum infrared thermography for detection of mary aim of this study was to test the association of infrared thermography (IRT) from
M2 digital dermatitis lesions on hind feet of
unwashed hind feet with painful M2 lesions under farm conditions, with lesion detection as
standing dairy cattle. PLoS ONE 18(1): e0280098.
https://doi.org/10.1371/journal.pone.0280098 ultimate goal. Secondary objectives were to determine the association between IRT from
washed feet and M2 lesions, and between IRT from unwashed and washed feet and the
Editor: Julio Cesar de Souza, Universidade Federal
de Mato Grosso do Sul, BRAZIL presence of any DD lesion. A total of 641 hind feet were given an M-score and IRT images
of the plantar pastern were captured. Multivariable logistic regression analyses were done
Received: April 29, 2022
with DD status as dependent variable and maximum infrared temperature (IRTmax), lower
Accepted: December 20, 2022
leg cleanliness score and locomotion score as independent variables, and farm as fixed
Published: January 17, 2023 effect. To further our understanding of IRTmax within DD status, we divided IRTmax into
Peer Review History: PLOS recognizes the two groups over the median value of IRTmax in the datasets of unwashed and washed feet,
benefits of transparency in the peer review respectively, and repeated the multivariable logistic regression analyses. Higher IRTmax
process; therefore, we enable the publication of
from unwashed hind feet were associated with M2 lesions or DD lesions, in comparison with
all of the content of peer review and author
responses alongside final, published articles. The feet without an M2 lesion or without DD, adjusted odds ratio 1.6 (95% CI 1.2–2.2) and 1.1
editorial history of this article is available here: (95% CI 1.1–1.2), respectively. Washing of the feet resulted in similar associations. Dichoto-
https://doi.org/10.1371/journal.pone.0280098 mization of IRTmax substantially enlarged the 95% CI for the association with feet with M2
Copyright: © 2023 Vanhoudt et al. This is an open lesions indicating that the association becomes less reliable. This makes it unlikely that IRT-
access article distributed under the terms of the max alone can be used for automated detection of feet with an M2 lesion. However, IRTmax
Creative Commons Attribution License, which
can have a role in identifying feet at-risk for compromised foot health that need further exam-
permits unrestricted use, distribution, and
reproduction in any medium, provided the original ination, and could therefore function as a tool aiding in the automated monitoring of foot
author and source are credited. health on dairy herds.
Data Availability Statement: Due to a signed
agreement between the researchers and farmers
we are unable to make all data underlying the
findings of this manuscript fully available without
existing M4 lesion [18], were not scored as such, and therefore lesions of this description were
included within the M1 category. The LS considered five classes, with 1 = perfect gait and
5 = severely lame, based on the 7 specific gait attributes as described by Flower and Weary [19]
and validated for use on video recordings by Chapinal et al. [20] and Ito et al. [21].
Statistical analyses
Statistical analyses were conducted using RStudio Version 1.3.1093 [23, 24]. Statistical signifi-
cance was declared at p < 0.05. Handling of the collected data for analysis of the different
objectives is detailed in Fig 2.
First, descriptive analyses were done to identify the number of feet with M2 lesions with
IRTmax available after software processing of the IRT images. At the first visit, 21 hind feet
met these requirements in the unwashed and washed condition. Another 15 unwashed and 19
Fig 1. Example of infrared thermography data collection and analysis of images from FLiR i3 handheld camera using ThermaCAM Researcher
Professional 2.8 SR-2 software.
https://doi.org/10.1371/journal.pone.0280098.g001
Fig 2. Study flow diagram for a study testing the association between broad spectrum infrared thermography and
the presence of digital dermatitis lesions using unwashed and washed hind feet from five Canadian dairy herds.
The first four herds were visited at 3-week intervals for a total of 12 weeks, resulting in five visits with data collection
per farm and the fifth farm was visited once. The bold lines represent general study recruitment, the solid lines
represent the unwashed hind feet and the dotted lines represent the washed hind feet. The diagram was created with
www.app.diagrams.net.
https://doi.org/10.1371/journal.pone.0280098.g002
washed hind feet were available from the other visits. Because of the low prevalence of M2
lesions in the dataset, it was decided to complement the data from the first visit with M2 scored
feet only from the following visits for further statistical analyses.
Results
A total of 641 hind feet from 310 cows of 5 farms were enrolled in the study (Fig 2). After dis-
carding feet missing an IRTmax value, either due to absence of an IRT image or inability to
process the IRT image with the software, and discarding feet missing an M-score, a total of 529
unwashed hind feet from 285 cows and a total of 558 washed hind feet from 289 cows with an
IRTmax value and an M-score were available for analysis. The unwashed dataset had 54 cows
with one observation, 218 cows with two observations, and 13 cows with three observations
with IRTmax and M-score data, whereas the washed dataset had 32 cows with one observation,
245 cows with two observations, and 12 cows with three observations with IRTmax and M-
score data. From these, 205 unwashed hind feet from 115 cows and 229 washed hind feet from
123 cows also had both LS and CS data available.
Lactating herd size ranged from 166 to 279 cows and farm-level DD prevalence (at least one
hind foot with DD) in enrolled cows ranged from 62 to 85% (mean 72%, standard deviation 9)
at the first visit. An overview of the M-scores by farm, LS, and CS for the hind feet with an IRT-
max in our study is provided in Table 1. The unwashed hind feet dataset contained 36 feet
with an M2 lesion and 493 feet without an M2 lesion, and 310 feet with DD and 219 feet with-
out DD. The washed hind feet dataset contained 40 feet with an M2 lesion and 518 feet without
an M2 lesion, and 329 feet with DD and 229 feet without DD. Table 2 provides an overview of
the descriptive statistics of IRTmax for each group of hind feet and boxplots of the IRTmax are
provided in Fig 3.
Table 1. M-scores for digital dermatitis, locomotion score, and cleanliness score for hind feet before and after washing from five Canadian dairy herds.
Unwashed hind feet Washed hind feet
M0 M1 M2 M3 M4 Total M0 M1 M2 M3 M4 Total
Herd
1 22 0 2 21 15 60 32 0 3 24 18 77
2 27 1 14 50 21 113 23 1 13 51 20 108
3 36 1 7 13 15 72 36 1 9 15 17 78
4 15 0 1 28 4 48 14 0 3 32 5 54
5 119 2 12 33 70 236 124 2 12 34 69 241
Total 219 4 36 145 125 529 229 4 40 156 129 558
LSb
1 45 1 10 57 20 133 49 1 12 60 23 145
2 10 0 6 20 7 43 11 0 6 19 8 44
3 7 0 0 8 6 21 11 0 2 11 8 32
4 4 0 2 2 0 8 4 0 2 3 0 9
5 0 0 0 0 1 1 0 0 0 0 0 0
Total 66 1 18 87 34 206 75 1 22 93 39 230
CSc
1 10 0 2 8 6 26 6 0 2 10 6 24
2 61 2 7 66 29 165 65 2 6 73 32 178
3 26 0 14 34 20 94 32 0 19 36 22 109
4 3 0 0 4 0 7 2 0 0 3 0 5
Total 100 2 23 112 55 292 105 2 27 122 60 316
a
M-stages [17] were determined in-parlour, after washing the feet with water. The M4.1 stage by Berry et al. [18] is included in the M1 stage.
b
Locomotion scores [19] were determined from video recordings of cows exiting the milking parlour with score � 3 considered lame; only available for feet from farm
1 to 4.
c
Lower leg cleanliness scores [15, 16] were determined in-parlour with presence of dried manure in score � 3; only available for feet from farm 1 to 4.
https://doi.org/10.1371/journal.pone.0280098.t001
Table 2. Descriptive statistics for maximum infrared temperature (˚C) of the plantar pastern from standing dairy cattle feet before and after washing, categorized
by digital dermatitis (DD) status.
DD status N mean SD minimum Q1 median Q3 maximum
Unwashed hind feet
M2a 36 32.1 1.2 28.8 31.5 32.2 33.0 34.3
M0|M1|M3|M4a 493 30.3 2.8 18.4 29.0 30.7 32.2 35.4
DD present 310 30.9 2.4 20.3 29.7 31.2 32.5 35.4
DD absent 219 29.7 3.1 18.4 28.1 30.3 31.9 34.4
Washed hind feet
M2a 40 32.1 1.3 29.0 31.3 32.3 32.9 34.7
M0|M1|M3|M4a 518 30.5 3.0 17.3 29.3 31.2 32.5 35.3
DD present 329 31.1 2.4 19.0 29.8 31.5 32.8 35.3
DD absent 229 29.9 3.4 17.3 28.4 31.0 32.4 35.1
a
M-stages [17] were determined in-parlour, after washing the feet with water; the M4.1 stage by Berry et al. [18] is included in the M1 stage.
https://doi.org/10.1371/journal.pone.0280098.t002
Fig 3. Boxplots for maximum infrared temperature (IRTmax) from the pastern region of dairy cattle hind feet
before and after washing. (A) For hind feet with M2 or M0|M1|M3|M4 lesions of digital dermatitis (DD). (B) For
hind feet with absence or presence of DD. Bold solid line = median, box = interquartile range (IQR), bottom
whisker = 25th percentile—1.5 x IQR, top whisker = 75th percentile + 1.5 x IQR, circle = datapoint outside the
interwhisker range.
https://doi.org/10.1371/journal.pone.0280098.g003
and washed hind feet (adjusted OR 1.1; 95% CI 1.1–1.2; Table 4). This association disappeared
after dichotomization of IRTmax.
Discussion
This multi-farm study provides insights into the practical application of IRT for detection of
DD, M2 lesions in particular, on hind feet from standing cows with a handheld IRT camera.
Higher IRTmax values were associated with feet with M2 lesions or DD lesions, in comparison
with feet without an M2 lesion or without DD, respectively, regardless of the hygienic condi-
tion of the feet. Dichotomization of IRTmax substantially enlarged the 95% CI for the associa-
tion with feet with M2 lesions indicating that the association becomes less reliable. When
looking at feet with any DD lesions, there was no association with the dichotomized IRTmax.
Previous work reported poor test characteristics to diagnose the presence of DD lesions using
IRTmax with Se 0.75–0.89 and Sp 0.65–0.70 [13, 29]. Altogether, these findings suggest that it
is unlikely that a cut-off value for IRTmax with high Se and Sp for the detection of feet with
M2 lesions can be determined using cross-sectional data.
Table 3. Final reduced multivariable logistic regression models to test the association between maximum infrared temperature (IRTmax) from the plantar pastern
and presence of M2 lesions [17] of digital dermatitis on hind feet from standing dairy cattle before and after washing, with lower leg cleanliness score (CS) [15, 16]
as explanatory variable and farm as fixed effect.
Model Variable Adjusted OR 95% CI
Unwashed hind feet
Continuous IRTmax
IRTmax + CS + farm IRTmax 1.6 1.2–2.2
CS fresh manure 1 referent
CS dried manure 4.1 1.6–10.7
Farm 1 1 referent
Farm 2 2.1 0.5–14.2
Farm 3 5.5 1.2–40.5
Farm 4 0.4 0.1–4.9
Dichotomized IRTmaxa
IRTmax + CS + farm IRTmax < 31.0˚C 1 referent
IRTmax � 31.0˚C 13.9 3.4–95.7
CS fresh manure 1 referent
CS dried manure 4.0 1.6–10.8
Farm 1 1 referent
Farm 2 2.2 0.5–15.2
Farm 3 6.4 1.3–48.7
Farm 4 0.5 0.1–5.6
Washed hind feet
Continuous IRTmax
IRTmax + CS + farm IRTmax 1.4 1.1–1.7
CS fresh manure 1 referent
CS dried manure 5.3 2.2–14.1
Farm 1 1 referent
Farm 2 3.9 1.1–17.9
Farm 3 10.7 2.7–56.0
Farm 4 1.9 0.3–11.4
Dichotomized IRTmaxa
IRTmax + CS + farm IRTmax < 31.3˚C 1 referent
IRTmax � 31.3˚C 4.8 1.7–15.8
CS fresh manure 1 referent
CS dried manure 5.5 2.3–14.5
Farm 1 1 referent
Farm 2 4.0 1.2–18.5
Farm 3 9.4 2.4–48.4
Farm 4 2.0 0.3–11.4
a
IRTmax was divided into 2 groups over the median value of IRTmax, regardless of M-score [17], in the datasets of unwashed and washed feet, respectively.
https://doi.org/10.1371/journal.pone.0280098.t003
In analogy with machine learning techniques used for automated mastitis or oestrus detec-
tion [30, 31], similar techniques can be developed to use IRTmax for the detection of M2
lesions in which the IRTmax from a foot is compared with rolling averages of the same foot,
contralateral foot, feet average within cow, herd average, or a combination of these. To date,
the authors are unaware of publications that report investigations of this option.
A limitation of this study was the low prevalence of feet with M2 lesions and of lame feet.
Although this is a realistic reflection of the average Canadian dairy herd [27, 32], it resulted
Table 4. Final reduced multivariable logistic regression analyses to test the association between maximum infrared temperature (IRTmax) from the plantar pastern
and presence of any lesions of digital dermatitis on hind feet from standing dairy cattle before and after washing, with farm as fixed effect.
Model Variable Adjusted OR 95% CI
Unwashed hind feet
Continuous IRTmax
IRTmax + farm IRTmax 1.1 1.1–1.2
Farm 1 1 referent
Farm 2 1.4 0.7–2.8
Farm 3 0.7 0.3–1.4
Farm 4 1.2 0.6–2.8
Farm 5 0.6 0.3–0.9
Dichotomized IRTmaxa
IRTmax + farm IRTmax < 31.0˚C 1 referent
IRTmax � 31.0˚C 1.4 0.9–2.1
Farm 1 1 referent
Farm 2 1.6 0.8–3.2
Farm 3 0.6 0.3–1.2
Farm 4 1.3 0.6–2.9
Farm 5 0.6 0.3–0.9
Washed hind feet
Continuous IRTmax
IRTmax + farm IRTmax 1.1 1.1–1.2
Farm 1 1 referent
Farm 2 2.7 1.4–5.2
Farm 3 1.3 0.7–2.7
Farm 4 2.4 1.1–5.4
Farm 5 0.8 0.5–1.4
Dichotomized IRTmaxa
IRTmax + farm IRTmax < 31.3˚C 1 referent
IRTmax � 31.3˚C 1.2 0.8–1.7
Farm 1 1 referent
Farm 2 2.6 1.4–5.1
Farm 3 0.9 0.5–1.7
Farm 4 2.1 1.0–4.7
Farm 5 0.7 0.4–1.2
a
IRTmax was divided into two groups over the median value of IRT max, regardless of M-score [17], in the datasets of unwashed and washed feet, respectively.
https://doi.org/10.1371/journal.pone.0280098.t004
in a statistically unbalanced dataset. It is possible that this restricted the capacity of our
study to detect an association with IRTmax. Also, our dataset contained a large number of
animals with only one observation, making the inclusion of cow as a random effect, to
account for repeated measures and cows having more than one observation, in our models
impossible.
The M-score of the feet in this study was determined by visual inspection of the feet in
the milking parlour. Although visual detection of M-scores in the milking parlour versus
in the trimming chute was validated [7], we hereby compared IRTmax with an imperfect
diagnostic test. Potentially, IRTmax could have correctly diagnosed some feet with M2
lesions that were misclassified as feet without M2 lesions by the in-parlour M-scoring
due to the limited Se (0.62) of in-parlour M-scoring for M2 lesions [7]. Diagnosis of DD
in the trimming chute would have reduced possible misclassification of M-scores. Addi-
tionally, information on the presence of other foot lesions, such as claw horn lesions,
could have been collected as feet with other foot lesions typically tend to have higher
IRTmax values compared to feet with no lesions [12, 33–35]. However, inspection of feet
in a trimming chute would have neglected the need for an easy, practical method. Higher
IRTmax values from cattle feet have also been associated with higher ambient tempera-
tures [33, 34], stage of lactation � 200 DIM [33], and more recently with higher locomo-
tion scores in a herd without DD [36]. Some of these factors will have been captured by
fixing farm into the models, but it is likely that they exert an unmeasured effect on the
results of our study.
Further research should aim to include all above-mentioned factors with a preference for
longitudinal studies to better evaluate IRT as an early detection method for M2 lesions result-
ing in lameness. However, these multiple factors which influence the ability to detect M2
lesions, and foot lesions in general, all need to be automatically measured and considered
before IRT can be easily used as a detection tool on farm. Until this further research is done,
the main potential use of IRT in automated detection of foot health status is likely limited to
identify ‘feet at risk’ that need further attention. At-risk feet could either be visually appraised
in the trimming chute, or by computer vision and machine learning technology. The YOLOv2
computer vision model of Cernek et al. [37] correctly classified about 60% of the lesions as an
M2 lesion on washed hind feet in an external validation trial on a commercial US dairy herd.
Combining IRT with other automated lameness detection devices presumably aids in the iden-
tification of feet at risk of compromised foot health.
Conclusions
The presence of M2 lesions on hind feet was associated with higher IRTmax values of the plan-
tar pastern, both on unwashed and washed feet from standing dairy cattle. Dichotomization of
IRTmax substantially decreased the reliability of this association, making it unlikely that IRT-
max alone can be used for automated detection of feet with an M2 lesion. It is probable that
IRTmax does have a role in identifying feet at-risk for compromised foot health that need fur-
ther checking and thereby is a tool that can aid in the automation of monitoring the foot health
status on dairy herds.
Supporting information
S1 Table. Technical data for the FLiR i3 handheld infrared thermography camera.
(DOCX)
S1 File. Regression analyses M2 lesions. Full results of all regression analyses performed
investigating the association between maximum infrared temperature (IRTmax) of the plantar
pastern region of feet from standing dairy cattle and the presence of M2 lesions of digital der-
matitis.
(DOCX)
S2 File. Regression analyses digital dermatitis lesions. Full results of all regression analyses
performed investigating the association between maximum infrared temperature (IRTmax) of
the plantar pastern region of feet from standing dairy cattle and the presence of any lesions of
digital dermatitis.
(DOCX)
Acknowledgments
The authors thank participating farmers for their willingness, time, and cooperation on this
project. Additional thanks to Kelsey Gray and Gwen Roy (University of Calgary, Alberta, Can-
ada) for assistance with data collection. Special thank you to Al Schaefer and Pierre Lepage
(Lacombe Research and Development Centre, Alberta, Canada) for assistance and guidance
regarding the use of IRT cameras. The authors are also grateful for the advice of Hans Vernooij
(Utrecht University, Utrecht, the Netherlands) on the statistical analyses and the help of
Lisanne van der Voort (Utrecht University, Utrecht, the Netherlands) for preparing the
figures.
Author Contributions
Conceptualization: Arne Vanhoudt, Casey Jacobs, Herman W. Barkema, Mirjam Nielen,
Tine van Werven, Karin Orsel.
Data curation: Casey Jacobs, Maaike Caron.
Formal analysis: Arne Vanhoudt, Casey Jacobs, Maaike Caron.
Funding acquisition: Casey Jacobs, Herman W. Barkema, Karin Orsel.
Investigation: Arne Vanhoudt, Casey Jacobs, Maaike Caron, Mirjam Nielen,
Tine van Werven, Karin Orsel.
Methodology: Arne Vanhoudt, Casey Jacobs, Mirjam Nielen, Tine van Werven, Karin Orsel.
Project administration: Casey Jacobs, Karin Orsel.
Resources: Casey Jacobs, Herman W. Barkema, Mirjam Nielen, Tine van Werven,
Karin Orsel.
Software: Arne Vanhoudt, Casey Jacobs, Maaike Caron.
Supervision: Casey Jacobs, Herman W. Barkema, Mirjam Nielen, Tine van Werven,
Karin Orsel.
Validation: Arne Vanhoudt, Casey Jacobs, Herman W. Barkema, Mirjam Nielen, Tine van
Werven, Karin Orsel.
Visualization: Arne Vanhoudt, Casey Jacobs, Maaike Caron.
Writing – original draft: Arne Vanhoudt, Casey Jacobs, Maaike Caron.
Writing – review & editing: Casey Jacobs, Herman W. Barkema, Mirjam Nielen, Tine van
Werven, Karin Orsel.
References
1. Orsel K, Plummer P, Shearer J, De Buck J, Carter SD, Guatteo R, et al. Missing pieces of the puzzle to
effectively control digital dermatitis. Transbound Emerg Dis. 2018; 65: 186–198. https://doi.org/10.
1111/tbed.12729 PMID: 29124910
2. Bruijnis MRN, Beerda B, Hogeveen H, Stassen EN. Assessing the welfare impact of foot disorders in
dairy cattle by a modeling approach. animal. 2012; 6: 962–970. https://doi.org/10.1017/
S1751731111002606 PMID: 22558967
3. Higginson Cutler JH, Cramer G, Walter JJ, Millman ST, Kelton DF. Randomized clinical trial of tetracy-
cline hydrochloride bandage and paste treatments for resolution of lesions and pain associated with dig-
ital dermatitis in dairy cattle. J Dairy Sci. 2013; 96: 7550–7557. https://doi.org/10.3168/jds.2012-6384
PMID: 24140336
4. Dolecheck K, Bewley J. Animal board invited review: Dairy cow lameness expenditures, losses and
total cost. Animal. Cambridge University Press; 2018. pp. 1462–1474. https://doi.org/10.1017/
S1751731118000575 PMID: 29557318
5. Döpfer D, Bonino Morlán J. The paradox of modern animal husbandry and lameness. Veterinary Jour-
nal. 2008. pp. 153–154. https://doi.org/10.1016/j.tvjl.2007.02.013 PMID: 17433736
6. Zinicola M, Lima F, Lima S, Machado V, Gomez M, Döpfer D, et al. Altered microbiomes in bovine digital
dermatitis lesions, and the gut as a pathogen reservoir. Guan LL, editor. PLoS One. 2015; 10:
e0120504. https://doi.org/10.1371/journal.pone.0120504 PMID: 25781328
7. Solano L, Barkema HW, Jacobs C, Orsel K. Validation of the M-stage scoring system for digital dermati-
tis on dairy cows in the milking parlor. J Dairy Sci. 2017; 100: 1592–1603. https://doi.org/10.3168/jds.
2016-11365 PMID: 27889123
8. Cramer G, Winders T, Solano L, Kleinschmit DH. Evaluation of agreement among digital dermatitis
scoring methods in the milking parlor, pen, and hoof trimming chute. J Dairy Sci. 2018; 101: 2406–2414.
https://doi.org/10.3168/jds.2017-13755 PMID: 29290450
9. Relun A, Guatteo R, Roussel P, Bareille N. A simple method to score digital dermatitis in dairy cows in
the milking parlor. J Dairy Sci. 2011; 94: 5424–5434. https://doi.org/10.3168/jds.2010-4054 PMID:
22032365
10. Stokes JE, Leach KA, Main DCJ, Whay HR. The reliability of detecting digital dermatitis in the milking
parlour. Vet J. 2012; 193: 679–684. https://doi.org/10.1016/j.tvjl.2012.06.053 PMID: 22874893
11. Mota-Rojas D, Pereira AMF, Wang D, Martı́nez-Burnes J, Ghezzi M, Hernández-Avalos I, et al. Clinical
Applications and Factors Involved in Validating Thermal Windows Used in Infrared Thermography in
Cattle and River Buffalo to Assess Health and Productivity. Anim 2021, Vol 11, Page 2247. 2021;11:
2247. https://doi.org/10.3390/ani11082247 PMID: 34438705
12. Stokes JE, Leach KA, Main DCJ, Whay HR. An investigation into the use of infrared thermography
(IRT) as a rapid diagnostic tool for foot lesions in dairy cattle. Vet J. 2012; 193: 674–678. https://doi.org/
10.1016/j.tvjl.2012.06.052 PMID: 22867853
13. Alsaaod M, Syring C, Dietrich J, Doherr MG, Gujan T, Steiner A. A field trial of infrared thermography as
a non-invasive diagnostic tool for early detection of digital dermatitis in dairy cows. Vet J. 2014; 199:
281–285. https://doi.org/10.1016/j.tvjl.2013.11.028 PMID: 24389042
14. Jacobs C, Orsel K, Mason S, Gray K, Barkema HW. Comparison of the efficacy of a commercial foot-
bath product with copper sulfate for the control of digital dermatitis. J Dairy Sci. 2017; 100: 5628–5641.
https://doi.org/10.3168/jds.2016-12257 PMID: 28478000
15. Cook NB. Hygiene Scoring Color Chart. 2006 [cited 22 Jul 2020]. Available: https://www.vetmed.wisc.
edu/fapm/wp-content/uploads/2020/01/hygiene.pdf.
16. Solano L, Barkema HW, Pajor EA, Mason S, LeBlanc SJ, Zaffino Heyerhoff JC, et al. Prevalence of
lameness and associated risk factors in Canadian Holstein-Friesian cows housed in freestall barns.
J Dairy Sci. 2015; 98: 6978–6991. https://doi.org/10.3168/jds.2015-9652 PMID: 26254526
17. Döpfer D, Koopmans A, Meijer FA, Szakáll I, Schukken YH, Klee W, et al. Histological and bacteriologi-
cal evaluation of digital dermatitis in cattle, with special reference to spirochaetes and Campylobacter
faecalis. Vet Rec. 1997; 140: 620–623. https://doi.org/10.1136/vr.140.24.620 PMID: 9228692
18. Berry SL, Read DH, Famula TR, Mongini A, Döpfer D. Long-term observations on the dynamics of
bovine digital dermatitis lesions on a California dairy after topical treatment with lincomycin HCl. Vet J.
2012; 193: 654–658. https://doi.org/10.1016/j.tvjl.2012.06.048 PMID: 22892182
19. Flower FC, Weary DM. Effect of Hoof Pathologies on Subjective Assessments of Dairy Cow Gait.
J Dairy Sci. 2006; 89: 139–146. https://doi.org/10.3168/jds.S0022-0302(06)72077-X PMID: 16357276
20. Chapinal N, de Passillé AM, Weary DM, von Keyserlingk MAG, Rushen J. Using gait score, walking
speed, and lying behavior to detect hoof lesions in dairy cows. J Dairy Sci. 2009; 92: 4365–4374.
https://doi.org/10.3168/jds.2009-2115 PMID: 19700696
21. Ito K, von Keyserlingk MAG, LeBlanc SJ, Weary DM. Lying behavior as an indicator of lameness in
dairy cows. J Dairy Sci. 2010; 93: 3553–3560. https://doi.org/10.3168/jds.2009-2951 PMID: 20655423
22. Harris-Bridge G, Young L, Handel I, Farish M, Mason C, Mitchell MA, et al. The use of infrared thermog-
raphy for detecting digital dermatitis in dairy cattle: What is the best measure of temperature and foot
location to use? Vet J. 2018; 237: 26–33. https://doi.org/10.1016/j.tvjl.2018.05.008 PMID: 30089541
23. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL https://www.R-project.org/. Vienna, Austria: R Foundation for Statisti-
cal Computing.; 2021. Available: https://www.r-project.org/.
24. RStudio Team. RStudio: Integrated Development for R. RStudio, PBC, Boston, MA URL http://www.
rstudio.com/. 2020.
25. Relun A, Lehebel A, Bruggink M, Bareille N, Guatteo R. Estimation of the relative impact of treatment
and herd management practices on prevention of digital dermatitis in French dairy herds. Prev Vet Med.
2013; 110: 558–562. https://doi.org/10.1016/j.prevetmed.2012.12.015 PMID: 23369719
26. Solano L, Barkema HW, Pajor EA, Mason S, LeBlanc SJ, Nash CGR, et al. Associations between lying
behavior and lameness in Canadian Holstein-Friesian cows housed in freestall barns. J Dairy Sci. 2016;
99: 2086–2101. https://doi.org/10.3168/jds.2015-10336 PMID: 26805982
27. van Huyssteen M, Barkema HW, Mason S, Orsel K. Association between lameness risk assessment
and lameness and foot lesion prevalence on dairy farms in Alberta, Canada. J Dairy Sci. 2020; 103:
11750–11761. https://doi.org/10.3168/jds.2019-17819 PMID: 32981721
28. Dohoo I, Martin W, Stryhn H. Veterinary epidemiologic research. Third Edit. Charlottetown, Canada:
VER; 2014.
29. Anagnostopoulos A, Barden M, Tulloch J, Williams K, Griffiths B, Bedford C, et al. A study on the use of
thermal imaging as a diagnostic tool for the detection of digital dermatitis in dairy cattle. J Dairy Sci.
2021; 104: 10194–10202. https://doi.org/10.3168/jds.2021-20178 PMID: 34099304
30. Reith S, Hoy S. Review: Behavioral signs of estrus and the potential of fully automated systems for
detection of estrus in dairy cattle. Animal. 2018; 12: 398–407. https://doi.org/10.1017/
S1751731117001975 PMID: 28807076
31. Steele NM, Dicke A, De Vries A, Lacy-Hulbert SJ, Liebe D, White RR, et al. Identifying gram-negative
and gram-positive clinical mastitis using daily milk component and behavioral sensor data. J Dairy Sci.
2020; 103: 2602–2614. https://doi.org/10.3168/jds.2019-16742 PMID: 31882223
32. Jacobs C, Orsel K, Mason S, Barkema HW. Comparison of effects of routine topical treatments in the
milking parlor on digital dermatitis lesions. J Dairy Sci. 2018; 101: 5255–5266. https://doi.org/10.3168/
jds.2017-13984 PMID: 29573803
33. Alsaaod M, Büscher W. Detection of hoof lesions using digital infrared thermography in dairy cows.
J Dairy Sci. 2012; 95: 735–742. https://doi.org/10.3168/jds.2011-4762 PMID: 22281338
34. Wood S, Lin Y, Knowles TG, Main DCJ. Infrared thermometry for lesion monitoring in cattle lameness.
Vet Rec. 2015; 176: 308–308. https://doi.org/10.1136/vr.102571 PMID: 25467147
35. Main DCJ, Stokes JE, Reader JD, Whay HR. Detecting hoof lesions in dairy cattle using a hand-held
thermometer. Vet Rec. 2012; 171: 504. https://doi.org/10.1136/vr.100533 PMID: 22961294
36. Werema CW, Laven L, Mueller K, Laven R. Evaluating Alternatives to Locomotion Scoring for Lame-
ness Detection in Pasture-Based Dairy Cows in New Zealand: Infra-Red Thermography. Anim an open
access J from MDPI. 2021; 11: 3473. https://doi.org/10.3390/ani11123473 PMID: 34944250
37. Cernek P, Bollig N, Anklam K, Döpfer D. Hot topic: Detecting digital dermatitis with computer vision.
J Dairy Sci. 2020; 103: 9110–9115. https://doi.org/10.3168/jds.2019-17478 PMID: 32861492